I know and took it as such without the need for tagging. But others seemed to not have gotten the message, ;-)
I tagged it /s for sarcasm, and even added a wink
Trading with Chat GPT
- Thread starter traider
- Start date
There has been a lot of talk about using AI for trading - personally - I have some doubts - at least at this early stage of development. I fully understand why traders - that are looking for help - are attracted.
I developed what I now call my DirtyAI - not truly AI - but my version of what it might be someday - when if comes to trading. I have been using it for a long time now - way before AI became the new popular thing. Will - AI - make it better - perhaps - but right now - I am staying with my version - my DirtyAI.
There is absolutely - NOTHING - to BackTest. By the way - in the attached pictures - the Algo is looking at the ES - NQ - RTY(Russell) and YM - at the same time - to make a decision on whether to Enter a trade or NOT. The results are for 10 days - 24/7. I can tell the Algo to trade for 1 hour - 2 hours - 15 hours - etc. - and it will stop trading - and Exit any trade that is open at the time. The Signals show 2 Contracts being traded - however - both started out with 1 Contract - 10 days ago - and eventually started trading 2 contracts. If there is a Big Loss - they could go back down to 1 Contract.
The second chart picture - shows what happened on the trade - as I was taking pictures. What I have - is NOT the Holy Grail - the Algo does LOSE - but that is what I call a "Controlled Loss". When I first came up with the Code - it kind of blew me away - and to myself - I did say I think I have the "HolyShit......Grail". It was kind of unbelievable to me - so I put the Code aside - didn't try to use it for a long time - as I couldn't believe the results - and to this day - some days - I still don't believe the results.
This is on MultiCharts PowerLanguage. The point of this - is to show traders - they do NOT necessarily need AI - as current software - NOT just MultiCharts - but MultiCharts.NET and NinjaTrader and possibly others - have the potential to do the same. I am currently working on code for MC.NET and NT as well.
As usual - I talk too much - LOL.
The point of all this - is a trader doesn't necessarily need AI.
Wish all traders the best.
Edit: Mark Brown - all the pictures are on Point (Range) Bars - but I noticed you also mentioned Kase Bars - so I changed it to Kase - does better than the Point Bars. I might look to see how Kase Bars work - compared to Point Bars. Also - the Algo can be put on any other type of bar - I just prefer Point Bars. Contact me - I will be glad to show you. I just remembered - I sent you a Private Message - a day or two ago - damn old age thing - totally forgot.
I developed what I now call my DirtyAI - not truly AI - but my version of what it might be someday - when if comes to trading. I have been using it for a long time now - way before AI became the new popular thing. Will - AI - make it better - perhaps - but right now - I am staying with my version - my DirtyAI.
There is absolutely - NOTHING - to BackTest. By the way - in the attached pictures - the Algo is looking at the ES - NQ - RTY(Russell) and YM - at the same time - to make a decision on whether to Enter a trade or NOT. The results are for 10 days - 24/7. I can tell the Algo to trade for 1 hour - 2 hours - 15 hours - etc. - and it will stop trading - and Exit any trade that is open at the time. The Signals show 2 Contracts being traded - however - both started out with 1 Contract - 10 days ago - and eventually started trading 2 contracts. If there is a Big Loss - they could go back down to 1 Contract.
The second chart picture - shows what happened on the trade - as I was taking pictures. What I have - is NOT the Holy Grail - the Algo does LOSE - but that is what I call a "Controlled Loss". When I first came up with the Code - it kind of blew me away - and to myself - I did say I think I have the "HolyShit......Grail". It was kind of unbelievable to me - so I put the Code aside - didn't try to use it for a long time - as I couldn't believe the results - and to this day - some days - I still don't believe the results.
This is on MultiCharts PowerLanguage. The point of this - is to show traders - they do NOT necessarily need AI - as current software - NOT just MultiCharts - but MultiCharts.NET and NinjaTrader and possibly others - have the potential to do the same. I am currently working on code for MC.NET and NT as well.
As usual - I talk too much - LOL.
The point of all this - is a trader doesn't necessarily need AI.
Wish all traders the best.
Edit: Mark Brown - all the pictures are on Point (Range) Bars - but I noticed you also mentioned Kase Bars - so I changed it to Kase - does better than the Point Bars. I might look to see how Kase Bars work - compared to Point Bars. Also - the Algo can be put on any other type of bar - I just prefer Point Bars. Contact me - I will be glad to show you. I just remembered - I sent you a Private Message - a day or two ago - damn old age thing - totally forgot.
Attachments
Last edited:
The reason for fake news is not to expect current readers to believe it but rather to create cited sources. Entire audit trails for fake news sources for algos created long before chat GPT. Chat GPT just takes available data and uses it. To provide a created, sourced historical narration to fit the needs of the writers.
News has never been so fake both MSN and financial news. Its ironic that Chat GPT has emerged at this time. It would make more sense to train neural networks to follow good traders within a specified market and timeframe and to ignore supposed tradeable news within the algo. To assume that most top trading houses have inside information and first hand news from the ground.
You want to also ignore 9am opening activity and end of day window dressing. Fed and governmental news is prolly the most fake. You want to train your AI program to strictly follow a trading floor of top traders with their respective markets and timeframes and clear all the white noise. Chat GPT will just trade index funds as wallstreet wants you to do. It will buy "after" the good quarterly reports ect.. Chat GPT relies on western news, english language. It will never give you clear decisions on current world events for trading. ChatGPT is a warm, fuzzy friend whom prolly will sooth your worried insecurities. But it will never be a trader.
News has never been so fake both MSN and financial news. Its ironic that Chat GPT has emerged at this time. It would make more sense to train neural networks to follow good traders within a specified market and timeframe and to ignore supposed tradeable news within the algo. To assume that most top trading houses have inside information and first hand news from the ground.
You want to also ignore 9am opening activity and end of day window dressing. Fed and governmental news is prolly the most fake. You want to train your AI program to strictly follow a trading floor of top traders with their respective markets and timeframes and clear all the white noise. Chat GPT will just trade index funds as wallstreet wants you to do. It will buy "after" the good quarterly reports ect.. Chat GPT relies on western news, english language. It will never give you clear decisions on current world events for trading. ChatGPT is a warm, fuzzy friend whom prolly will sooth your worried insecurities. But it will never be a trader.
Last edited:
Good morning VSTscalper,
Very good job on your algo!
Question please:
1. Why do you say "There is absolutely - NOTHING - to BackTest."
the coding aspect alone can replace an office staff of menza members lol
The picture shows an Algo I coded many years ago - for Scaling In and Scaling Out. It does NOT Scale Out - on all the trades - as it depends on what is happening in the market. And - if you look at the date - I am NOT sure if AI was around at that time.
I do understand - for many traders - why AI could potentially help with their trading - but again - the trading software we have today - still has a lot of potential.
I do understand - for many traders - why AI could potentially help with their trading - but again - the trading software we have today - still has a lot of potential.
Attachments
Last edited: